Research

My research is mostly about computational and ecological statistics. I am particularly interested in developing statistical models for complex ecological data and efficient model-fitting methods (Bayesian or frequentist) for complicated models (e.g., hierarchical models). I am also keen on developing general, efficient, and user-friendly open-source statistical software.

A lot of my previous work focused on probability density approximation methods and their applications in likelihood-based inference, for example, the saddlepoint approximation method and the maximum entropy principle. I was also interested in methods for addressing computational challenges involved in applying these approximate methods. For complicated ecological, biological, or genetic models, the likelihood function is often intractable: it might be computationally intensive to evaluate, unknown, or too complicated to derive analytically. These models may be fitted using the Bayesian MCMC methods or approximate Bayesian computation. However, model complexity has outstripped the pace of computational advances. I am interested in developing approximate methods that are both fast and accurate for fitting these models.